Modeling of Power Line Communications

for Indoor System Using MIMO coding

scheme
B.Rajkumarsingh and Noorianah Lollmahamod
Dept. of Electrical and Electronic Engineering
University of Mauritius
Rduit, Mauritius
email: b.rajkumarsingh@uom.ac.mu/noori_001@live.com

I. INTRODUCTION
Abstract With the advent of technology, human
dependency on electricity and communication has grown
beyond leaps and bounds. As a matter of fact, Power line
communication has been of remarkable interest in the past few
years due to the availability of power lines in every home. In
previous fields, narrowband Power Line Communications has
been employed for potential indoor as well as outdoor
applications. However, with the spread of broadband
technologies, the need to study broadband Power Line
Communications in indoor use is essential. Therefore, this
paper aims at modelling the broadband Power line
communications channel based on indoor applications as well
as applying Multiple Input Multiple Output coding technique
with its channel estimation to the system to achieve better
performance. The main objectives were to model the Power
Line communications channel and a performance analysis of
the Power line communications system is done under Rayleigh
Fading distribution along with its noise modelling. Assuming
BPSK modulation scheme, the system performance with
varying load impedances is compared. After applying MIMO
codes, performance comparison is analyzed with varying
number of transmit and receive antennas as well as code rates.
The results have shown that narrowband scenario has a better
performance than broadband scenario due to its lower
attenuation. Also, it has been seen that load impedances greatly
affect the BER performance and these are dependent on the
weighting factors. Similarly, the results have proved that PLC
channel follow a Rayleigh Fading distribution. Compared to
the SISO model, the 2x2 MIMO shows an improvement of 5.8
dB. With increasing receive antennas, the system performs
better. Comparison of the code rates has demonstrated that the
rate coded system achieves greater performance than the
rate coded system. These analyzed performances have shown
that MIMO coding scheme can be implemented in broadband
PLC channel for indoor applications.

Power line communications (PLC) is an ancient idea that

goes back to the early 1920s which meant that data could
be transferred through power lines. The power line
communication systems involve terminal devices that are
plugged into the electrical power supply network and allow
data to be transmitted through the network to other terminal
devices attached to the network. It has been used for
narrowband relay applications, that is, low data rates of up
to 500 Kbps, such as the remote control, public lightning and
domestic installations during the early years. It is only
during the end of 1990s that this technology has been
developed to allow higher transmission speed of data of up
to 200 Mbps known as broadband PLC. In fact, this
technology has brought about many advantages including
the idea of its availability all around the world through the
sockets in our home [1].
Due to its broad coverage and with no new infrastructure
needed, broadband over power lines has attracted much
attention to residential and commercial markets which
include both indoor and outdoor applications. With regards
to indoor applications, this technology is making it easier to
realize users entertainment needs such as HDTV, rough
control of smart appliances, the facility to distantly manage
electrical devices [2].
Standards have been set up in order the spectrum coverage
is respected and the system is controlled. The frequency
bands have been approved by the International
Telecommunications Union and they are generally
subdivided into narrowband (NB) and broadband (BB) PLC.
Most NB-PLC devices implement IEEE 1901.2, ITU.T
G.9904 and ITU.T G.9903 standards whereas BB-PLC
systems use the IEEE 1901.2010, ITU-T G.hn standards and
more recently, IEEE 1905.1 standard [3].

Index TermsPower Line Communications, Multiple Input

Multiple Output, Multipath Rayleigh Fading Channel, channel
estimation.

According to data transmission theory, it has been

analyzed that when data travels through a medium, it comes
upon numerous limitations such as multipath effect, fading,
attenuation and other sources of interference. As a matter of
fact, these restrictions need to be considered when modeling
the PLC channel as well as the PLC noise model which
greatly affects the system [3]. Thus, Section II will focus on
the modeling of PLC channel as proposed in [4] and based
on [5], PLC channel over Rayleigh fading channel model
will be analyzed. Compared to previous work carried out in
[22] whereby PLC channel is considered as a frequency flat
channel with STBC coding applied, this paper will deem the
PLC channel as frequency selective channel under Rayleigh
distribution using orthogonal STBC coding along with
channel estimation. Indoor applications shall be considered
together with the modeling of PLC noise approaches [3].
Section III will concentrate on the evaluation of the
multipath parameters based on the type of cables used for
indoor system. Section IV will deal the assessment of the
PLC noise parameters. Furthermore, the next section will
focus on the simulation results which comprise of the
attenuation and phase effects on the frequency and distance,
the performance comparison with the two channel models,
effects of varying load impedances, analysis of the
narrowband and broadband scenario, SISO and MIMO PLC
system whereby the latter has been used as the channel
encoder to improve the performance, effect of varying
transmit and receive antennas and code rates. The last
section will conclude the results obtained from the
implementation of the PLC channel.

= ZA and Z12 = ZC, thus reflection will occur only along the
segment BD and Zl3 ZD.

Figure 1: Multipath T-type topology

TA, TB, TC and TD represent the points forming the Ttopology.

The following equations are used to evaluate the

transmission and reflection coefficients.

= 11 23 ;
23 = 12 ||13
(1)
11 + 23

13 21
13 + 21

;
=

II. LITERATURE REVIEW

21 =
13

11 ||12

+ 13

= 1 +
= 1 +

A. Multipath effect in PLC

PLC channels are affected by multipath phenomenon like
wireless communication channels. This would mean that
signal propagation does not only take place through only a
line of sight path between transmitter and receiver but
additional echoes are also considered. This is due to the
several reflections initiated by junctions in house service
cables, house connection boxes, etc. with diverse
characteristic impedances [6].
In [7], it is approved that an indoor PLC network
frequently shows a T type topology. When a signal passes
through the tree from source to destination, the signal energy
will be divided by the branches at the joints, and echoed at
the branch ends due to impedance mismatches, resulting into
a grouping of the signal circulating through the direct path
from the transmitter and a number of reflections from
network branches. As seen in Figure 1, there are 3 segments
with lengths l1, l2 and l3 with their respective characteristic
impedances Zl1, Zl2 and Zl3.
Theoretically, it can be said that there are infinite number
of reflection paths and the outcome is a multipath channel
with frequency selective fading. For a basic consideration as
assumed in [8], TA and TC are matched which means that Zl1

(2)
(3)
(4)
(5)

Each transmitted signal reaches the receiver through N

different paths. Each path i is well-defined by a certain delay
i and a certain attenuation factor Ci. As such, the PLC
channel can be termed by means of a discrete-time impulse
response h (t). The impulse response of the channel h (t) can
be inscribed as an addition of the delayed and attenuated
Dirac pulses [9].

The transfer function leads to the equation as follows:

H (f) = =1 . (, ). 2

(6)

If the transfer function is to be expressed in terms of its

phase, it becomes:
() ()
H (f) =
.
=1 .

(7)

N is the number of fading paths, each path i has a

weighting factor gi which represents the product of the
transmission and reflection coefficients along the path and it
is usually less than 1. A (f, di) is attenuation factor caused by
2

PLC transmission cable losses in the form of heat or signal

leakage and f is the frequency of operation. The delay i of a
path i can be obtained from the length di, and the speed of
light in vacuum c0 and the relative permittivity r as below:
i =

0
Where, is the dielectric constant of the insulating
material.

equal variance and zero mean. Frequently, the gain and

phase components of the channels distortion are suitably
characterized as a complex number. Its probability density
function is given by:
2 /(2 2 )
f (x; ) = 2
, x 0 (11)

(8)

Where,
is the scale factor of the distribution.
PLC channel can be said to be frequency-selective. This
would mean that its channel spectral response contains fades
due to the numerous reflections causing elimination of
certain frequencies at the receiver or deep nulls due to
destructive interference [12].
Another important factor of Rayleigh fading channel is
the Doppler shift which is caused by motion of objects.
However, in the case of a PLC channel which is a wired
medium, there will no Doppler effect since there is no
movement of any object. Therefore, there will no Rayleigh
fading due to Doppler Effect but only due to multipath
propagation.

Based upon extensive study [10], A (f, di) can be

estimated by the mathematical formula for attenuation
factor . As such, (f) can be obtained from the sum of
attenuation parameters. It can be calculated using the
equation 2.8. This will lead to:
A (f, di) = ()
(9)
Where, (f) is the attenuation constant and is evaluated as
follows:

= + =

+
2

(10)

Table I shows the possible fading paths as well as their

weighting factors from transmitter to receiver.

C. PLC noise
In [13], it states that noise is one important feature which
influences the digital communications over power line
networks apart from signal distortion due to cable losses and
multipath propagation cited previously. Also, unlike other
communication channels, power line noise does not follow
the typical additive white Gaussian noise (AWGN) model.
It originates from electrical devices linked to the power grid
and exterior noise attached to the indoor network via
radiation or conduction. Based upon extensive study in [3],
it has been categorized into the following terms:

TABLE I: SIGNAL PROPAGATING PATHS

The multipath effect of the PLC is commonly used to

model the PLC channel, basically known as the echo-based
model. Further explanation is given in [4].

Impulsive noise: This consists of different components

that can also be classified according to:
Periodic impulsive noise synchronous with the
mains: It originates from controlled rectifiers in
power supplies operating synchronously with the
mains.
Periodic impulsive noise asynchronous with the
mains: It is generated from switched power
supplies and AC/DC power converters.
Asynchronous impulsive noise: It is caused by
switching transients initiated by connection and
disconnection of electrical appliances.
Narrowband Interference: This is usually caused by
modulated and sinusoidal signals originating from
broadcast stations, electrical appliances.
Background Noise: This comprises of the remaining
types of noise that are not counted in the previous
categories. It results from the summation of several
low power noise sources of indefinite origin.
Despite the short duration of impulse noise, it is
considered as the main noise source over the PLC channel.
The impulsive noise has been analyzed in both the frequency

B. Fading effect in PLC

From the above explanation, it is cleared that the behavior
of a power line channel is quite different from that of a
conventional wired communication channel due to its
unstable nature of its basic factors, that is, noise, delay and
attenuation. Nevertheless, the fading effects due to multipath
have been ignored in the derivations above. Fading
outcomes in degradation of the performance of a PLC
system as a whole and reduces the signal power but it has no
effect on the noise power. Several fading models have been
suggested in literature for the PLC channel.
In a more recent study conducted in [11], it has been
detailed that the Rayleigh fading model is extensively used
for modelling the PLC channel. It was experimentally
proved that parameters of the measured transfer functions
follow the Rayleigh distribution. As a matter of fact, this
paper will consider the Rayleigh fading effects to be valid
for the PLC channel using the parameters written above.
A channel is said to be Rayleigh-distributed if the real and
imaginary parts of the response are i.i.d., that is,
independently and identically distributed Gaussian with
3

and time domain. Using the time domain, impulsive noise

can be characterized by random variables namely the
impulse with twi, the impulse amplitude Ai and the interarrival time tarri for the ith impulse envelope as shown in
Figure 2. From experimental results carried out for indoor
broadband impulsive noise [14], their statistical properties
have been determined shown in Table II.

PLC modem (mains) and the TV set are assumed to be

constant, i.e. ZA = Z11, ZC = Z12 whereas the load impedance
of the microwave ZD is to be varied to create impedance
mismatching. Since the two appliances are linked to the
same type of cables, Z13 =Z12.

Figure 3: Indoor power line network topology

Typically, for indoor power line distribution, two-wire

transmission line are used and as such, NAYY35 and
NAYY150 cables are employed for outlet-outlet and interjunction connections respectively [12] the two-wire cables
are shown in Figure 4 below.

Figure 2: Impulse envelopes of impulsive noise

TABLE II: STATISTICAL PROPERTIES OF IMPULSIVE
NOISE

ai = radius of copper
conductor

ti = thickness of PVC
insulation: comparable with
that of the air space between
the conductors

Di = distance between
the two conductors

Narrowband interference can be modelled as the sum of

sinusoidal waveforms, denoted mathematically below [15]:

() =
=1 (). sin(2 + )

(12)

N is the number of carriers, each having a different

amplitude Ai(t), different phase and different frequency fi.
The phase can be selected randomly in the interval [0, 2]
and does not depend on time.

NOTE: i =1 for NAYY35; i =

3 for NAYY150.
Figure 4: Cable properties

From Figure 4, Di is obtained from the following equation:

Background noise is usually modelled in the frequency

domain and as stated in [6], power spectral density of
background noise in PLC channels is inversely proportional
to frequency. Several studies carried out have brought about
that the background noise can be considered to a sum of two
Gaussian distributions or Rayleigh distributions [16].

Di = (4.ti) + (2.ai)

(13)

The RLGC parameters are evaluated using the equations

from 14 to 17 [1]. However, in an actual transmission line,
attenuation caused by R and G is much less than by L and C,
i.e. R wL and G wC yielding a simplification as in
equation 18 [14].

III. EVALUATION OF BROADBAND INDOOR PLC CHANNEL

PARAMETERS

Indoor power line network utilizes the tree topology

network as illustrated in Figure 3. It actually comprises of
branches which are linked to a main path between the
transmitter and receiver. Individual branches represent
appliances with the appropriate load impedances. The figure
illustrates a PLC modem connected to the mains (A) with a
NAYY150 cable and this is linked to two sockets whereby
two appliances (TV set, Microwave) (C, D) are connected
by NAYY35 cables. B represents the mismatch point
whereby reflection will occur. The load impedances of the

Resistance, R =

Inductance, L =

Conductance, G =

(14)

Capacitance, C =

(15)

(16)

)
2

(17)

acosh( )
2

acosh(

Zc =

40 and 70 are taken for the analysis. This is illustrated

in Table V and as it can be observed, the load impedance
does not have any effect on the first path since it is not
reflected and thus no reflection coefficient. At 40 , the
weighting factors as from path number 4 can be neglected.
This is due to the matching properties of the impedance of
NAYY35 such that ZD Z12. However, the 10 and 70
impedances are different from Z12 resulting into a distortive
effect of the received signal. Compared to 70 , the 10
impedance will create a greater distorted signal due to its
negative echoed versions.

(18)

In case of a lossless transmission line, the phase constant

from equation 7 can be expressed as:

=
(19)

Where,
vp is the propagation velocity (m/s) and is equal to speed of
light/r
Where,
a = Radius of the conductor (m)
D = Distance between cables (m)
c = Permeability of conductor (H/m) = ro
r = Relative permeability of conductor
o = Free space permeability (H/m)
c = Conductivity of conductor (S/m)
d = Conductivity of dielectric material = d.tan
tan = Dissipation of dielectric material
1
= Skin depth (m) =

TABLE V: EVALUATED WEIGHTING FACTORS

cc

f = frequency of wave (Hz); angular frequency, w =

2f
d = permittivity of dielectric material (F/m) = ro
r = Relative permittivity of dielectric material
o = Free space permittivity (F/m)
Figure 5 shows the attenuation profile of each path from
Table IV with a frequency range of 30 MHz. As it can be
observed, the attenuation increases with increasing distance
and frequency. As studied in Literature Review, attenuation
is caused by cable losses. Thus, at higher frequencies, the
losses are more and cause reduction in the channel gain
meaning that BB-PLC suffers more attenuation than NBPLC. Moreover, at 20 MHz, with each propagating distance
from path 1 to path 6, the attenuation is seen to increment by
about 2.75 dB as cable losses are highly dependent on
propagating distance.

TABLE III: LINE PARAMETER VALUES

Table IV illustrates the lengths of the paths with their

respective path delays and path gains obtained from the
simulation. The path delay which denotes the time taken to
reach the receiver is in the order of 2. The results show that
with increasing path length, the path delay also increases,
roughly by two times. Also, the path gain in dB increases
approximately by 3 times. As observed, the path gains
values are negative in dB indicating that the power of each
path attenuates with increasing length.
TABLE IV: PATH LENGTHS WITH THEIR RESPECTIVE
DELAYS AND GAINS

Figure 5: Attenuation profile of each propagation path

The phase profile within a frequency range of 1 MHz for

each path is illustrated in Figure 6. As it can be perceived,
all the path signals follow a sawtooth wave, i.e. they ramp

In order to create impedance mismatching in the system,

the load impedance at node D, ZD is made to vary and its
effects on weighting factors are observed. Values of 10 ,
5

downwards to about -3 radians and then sharply rises to +3

radians. Nevertheless, as each path increases, the phase
shifts the signal at a lower frequency. As seen, the phase of
path 6 increases abruptly at around 220 kHz and path 5
around 260 kHz resulting in a difference of 40 kHz. This
difference increases as the length of path decreases. The
phase shift occurs due to the path delays and therefore, with
a greater path delay results into a phase shift at lower
frequency.

model. As matter of fact, it can be realized using an FIR filter

whereby the numerator coefficients represent the weighting
factors as proposed in [4].
However, in view of the fact that fading is a major
phenomenon that should be considered in a multipath
channel, the PLC channel can also be modelled using a
multipath channel with a Rayleigh fading distribution. This
has been extensively been studied in Section II and in [5].
Intrinsically, in Simulink, the MIMO multipath fading
channel block is utilized to implement the model whereby
the discrete path delays and average path gains coefficients
represent the path delays and path gains evaluated in III
shown in Table IV. The fading distribution is set to Rayleigh
and the Maximum Doppler Shift parameter is assigned to 0
since there is no Doppler effect in a wired system(refer to
Section II). For uncoded system fading model, an equalizer
may be necessary to be introduced to eliminate the ISI
caused by fading effects.
B. Modeling of PLC noise
Impulsive noise can be modelled from a set of pulse trains
with random amplitudes, pulse widths and phase delays as
per specifications and as such, pulse generators are used.
The period of the total pulse generators is determined using
the following equation [4]:

Figure 6: Phase profile for each propagation path

_
= =1
,

IV. COMPLETE PLC SYSTEM

(20)

The total impulsive power is derived from [4] and is shown

in equation 21.

MIMO coding scheme is applied to the system in order to

improve its performance and the overall diagram is shown
in Figure 7. The software MATLAB/Simulink R2014a is used
to realize the overall system and the corresponding
parameters are evaluated in MATLAB R2014a which are
then transferred to the required blocks.

_ =

_
2 . ,
=1

(21)

To generate narrowband interference, two sine wave

generators are used whereby random amplitudes and
frequencies are defined. The total sinusoidal power is
defined as follows:
= _
=1

2
2

(22)

The sum of two Gaussian noise generators are used to

model the background noise whereby their variances
represent the average background power and they are based
on the following assumption [17]:

Figure 7: Complete system model

1 2 = 102 2

All through the simulation, the data rate and the packet
size are taken to be 6 Mbps and 300 respectively. The BER
results are obtained after analyzing one million symbols. A
range of the SNR values were used to evaluate the BER. The
signal power is taken to be 0.5 W. BPSK method is used for
the simulation process before which, a buffer is added to
make the sample time of input same as output.

(23)

Background power is obtained from the subtraction of the

sum of impulsive power and sinusoidal power from total
noise power. Thus, for different values of SNR, the noise
power is varied.
C. MIMO coding scheme with channel estimation
PLC channel is affected by multipath and fading and thus
to combat these limitations, MIMO system can be applied.
In [18], it has been said that MIMO greatly helps in

A. Modeling of broadband indoor PLC channel

Based on the study carried out in II, it was said that the
PLC channel model can be assumed to be an echo-based
6

achieving higher data rates and lower BER. MIMO uses the
diversity coding technique whereby it uses multiple
antennas at both receiver and transmitter side to send
multiple copies of the transmitted signal which are
independently faded.
MIMO systems can be defined as a link in which the
transmitting and the receiving ends are equipped with
multiple antenna elements as illustrated in Figure 8. The core
idea behind the MIMO system is the Space Time Block
Codes (STBC) in which time is complemented with the
spatial dimension inherent in the use of multiple spatially
distributed antennas which means that data is coded with
space and time [19].

channel estimate from all the transmit antennas, so as to

support the channel to approximate about the signal to
extract the data encoded in the symbols [19].

Figure 9: Alamouti space time decoder

The ML detector selects a pair of signals (1, 2) from the

signal constellation to reduce the distance metric over all
possible values of 1 and 2, which means that it takes the
message yielding the least distance between the received
vector and theorized message [19].

Figure 8: Block diagram of MIMO system

= || ||2

The received vector is given by:

(25)

d2 (r1, h1 1 + h2 2) + d2 (r2, - h1 2 + h2 1*) = | r1 - h1 1 h2 2|2 + | r1 + h1 2 - h2 1*|2

(26)

y = Hx + n
(24)
Where,
x is the transmitted signal
H is the channel gain matrix and
n is the noise vector
11 12 1
H can be described as (21 22 2 ) with a Nt x Nr
1 2
dimension where each entry hi,j denotes the phase shift and
attenuation (transfer function) between the ith transmitter
and jth receiver. Space Time Block Codes are a general
version of Alamouti scheme [20]. These codes are
orthogonal and can attain full transmit diversity quantified
by the number of transmit antennas. The data are built as a
matrix which has its columns equivalent to the number of
transmit antennas and its rows equivalent to the number of
time slots necessary to transmit the data. At the receiver side,
the signals received are first combined and then sent to the
maximum likelihood detector where the decision guidelines
are employed [19]. The code rate which is denoted by the
number of transmitted symbols over the number of time slots
is 1 for the Alamouti scheme. It means that the Alamouti
scheme achieves full diversity gain without sacrificing its
data rate [20]. However, for higher number of transmit
antennas, full rate cannot be achieved and therefore it is
necessary to sacrifice data rate. As such, the code rate can be
or depending on the encoding algorithm [19].
The OSTBC decoder block has a simple decoding system
based on maximum likelihood at the receiver [21]. As shown
in Figure 9, the estimated coefficients of the channel, i.e. the
channel estimator and the combiner are given as input to the
ML detector. The channel estimator is a significant part of
the decoder whereby each transmitting and receiving
antenna characterize a channel coefficient. The decoder
accepts and combines all the input signals along with the

Where,
r = received vector
H = channel gain coefficient
d = distance metric

V. SIMULATION RESULTS
A. Performance of PLC channel (FIR filter) with varied
load impedances
The BER performance of the broadband indoor PLC
system is compared using three different scenarios by
varying the load impedances. As it can be observed in Figure
10, the 40 system shows a better performance than that of
10 and 70 . For the 70 system to reach the curve of 40
at BER of 10-3, an SNR of 1 dB is required. However, the
systems performance is improved as the load impedance
varies from 10 to 70 by 4.75 dB at a BER of 10-2. The
variation observed in the BER performance is due to the
evaluated weighting factors as explained in Section III.

Noticeably, the 2x2 reaches a BER of 10-4 at an SNR of 6

dB and the 2x3 and 2x4 with a steeper decrease at 5.5 dB
and 5 dB respectively, having an SNR of 0.5 dB difference.
Therefore, it can be said that the 2x4 system shows a better
performance than that of 2x2 and 2x3 systems, having a
greater number of receive antennas with an improved spacetime diversity.

Figure 10: Performance comparison with varied load impedances

B. Performance comparison with the two PLC channel

models proposed
As it can be observed in Figure 11, the systems
performance is improved with Rayleigh channel as
compared to the Digital filter model. At about a BER of 10 1
, there is a gap of 6 dB between the two curves. However,
as the SNR increases, the gap decreases and at BER of 10 -2,
the gap is seen to be about 0.5 dB. In other words, at lower
SNRs, there is a greater SNR difference than at higher SNRs.

Figure 12: Performance comparisons of varied receive antennas

with code rate of 1 with SISO model

D. Performance analysis with varying code rates and

antennas
As the number of transmit antennas becomes greater than
2, it can have a code rate of either or . As illustrated in
Figure 13, the systems performance is analyzed with a code
rate of . A significant improvement is observed in the 4x4
MIMO as compared to the 3x3 MIMO with an SNR decrease
of 5.5 dB at a BER of 10-4.
Comparing the BER curves with 4 transmit antennas, it
can be noticed that an SNR decrease of 1 dB occurs from
4x3 to 4x4 at BER of 10-5and 1.3 dB from 4x2 to 4x3. As
expected, the BER curves having 4 transmit antennas reach
the waterfall region quicker than that with 3 transmit
antennas. The latter seem to have a flattening curve which is
a less desirable performance.

Figure 11: Performance comparison with two PLC channel

models

C. Performance of indoor PLC system with MIMO codes

The improved trend of the MIMO coded system, from 2 to
3 receive antennas, with a code rate of 1 can be observed in
Figure 12. With the introduction of the Orthogonal STBC
codes in the system, the performance of the 2x4 MIMO is
improved by roughly 4.8 dB at a BER of 10-2, compared to
the SISO system. Besides, a significant decrease of about 5
dB is observed for the 2x2 and 2x3 system.
8

E. Performance comparison with previous work

The performance comparison between the implemented
system and that evaluated in [22] is realized in this section.
The previous work analyzes the PLC channel which is
modeled as a flat fading channel with 3x3 STBC codes using
BPSK modulation scheme. The system is severely degraded
by 12 dB, as clearly observed in Fig.15, compared to the
simulated one. The improvement in the implemented system
can be accounted by the use of OSTBC codes along with
channel estimation. However, this comparison cannot be
fully done as the PLC modeling is not similar since the
previous work implemented the PLC as frequency flat and
this paper considered PLC as frequency selective. Moreover,
as the previous work also made use of convolutional coding
and interleaving, their effects are quite impressive, with a
remarkable improvement of 3.8 dB compared to the
implemented system.

Figure 13: Performance comparisons with code rate

As far as the rate coded system is concerned, a visible

aspect is that the rate coded system enters the waterfall
region at relatively smaller SNR than the rate coded
system, meaning that it encounters a greater improvement at
the expense of lower data transmission. As observed in
Fig.14, the performance of 4x4 MIMO yields into an
advance of 5 dB compared to the 3x3 MIMO at a BER of
10-4.
Furthermore, with greater number of receive antennas, the
system performs better, improving by SNR 1 dB and further
1.3dB. An abrupt decrease of BER is observed for the 4x2
MIMO around 2 dB. For 3 transmit antennas, it can be seen
that they have a rather flattening BER curve and enter the
waterfall region at high SNRs. It can be deduced that the
rate coded system is less desirable due to its degradation
which may be accounted for the increase in transmitted
symbols and hence more prone to errors. However, relative
to SISO systems, the rate coded system shows a greater
progress [21].

Figure 15: Performance comparison with previous work

VI. CONCLUSION
The main objective of this paper was to analyze the
performance of the broadband indoor PLC channel using
two different channel models. Implementation of the
multipath PLC model along with its noise modelling were
done. The systems performance was put into comparison
with a PLC-MIMO coded system with varying antennas and
code rates. From the attenuation profile, it has been observed
that the attenuation rises as the distance and frequency
increase. It increments by about 2.75 dB as the propagating
distance increases from path 1 to path 6 at 20 MHz. Thus, it
can be concluded that BB-PLC suffers more attenuation than
NB-PLC. For the phase profile, it is seen that as distance
increases, the frequency at which the signal is shifted by
phase decreases. PLC. Furthermore, it has been observed
that the systems performance is greatly affected by load
impedances whereby mismatching occurs causing reflection
and this in turn depends on the weighting factors. As the
BER performance of different load impedances was
analyzed, it showed that the 10 system was degraded by

Figure 14: Performance comparisons with code rate 3/4

4.75 dB compared to the 70 at a BER of 10-2 as its factors

were extremely destructive and the 40 outperformed the
70 by 1 dB at BER of 10-3 since it involved less vicious
factors.
Two channel models, Digital FIR filter and Multipath
Rayleigh Fading Channel models were used to model the
PLC channel and their BER performance were investigated.
The FIR filter considered the weighting factors whereas the
Rayleigh fading channel considered the path delays and path
gains of the channel. The results obtained, showed that at
high SNRs, the BER curve of the Rayleigh model follows
almost that of the echo-based FIR filter model. Hence, it can
be said that PLC channel can be modelled as a multipath
channel with its paths having a Rayleigh distribution. For the
indoor PLC-MIMO system, it has been shown that
compared to the SISO model, a 2x4 MIMO undergoes an
improvement of 4.8 dB at BER of 10-2 and it improves by 1
dB compared to the 2x2 MIMO, indicating that the increase
in number of receive antennas improves the BER
performance. Upon comparing the code rates and varying
the number of transmit and receive antennas, it is observed
that there is a significant improvement of 5.5 dB of the 4x4
MIMO compared to the 3x3 MIMO at BER of 10-4 and the
rate coded system proves to be more remarkable than the
rate coded system. Comparison with previous work yields
an improvement of 12 dB, explaining the effect of channel
estimation and the use of orthogonal STBCs. With the use
of interleaver and convolutional coding in previous work, it
is seen that it outperforms the implemented system with an
improvement of 3.8 dB. Nonetheless, this comparison is not
much reliable as the PLC channel is not modeled similarly.
The need to combat the fading effects of the PLC channel
is vital and therefore, the use of equalizer and MIMO codes
prove to be powerful tools to reduce these anomalies. MIMO
coding system, along with its channel estimation improve
the multipath effect of the PLC channel and thus can be
implemented in indoor broadband systems where great
attenuation can be encountered.
For future work, a more efficient modulation scheme such
as OFDM can be implemented, along with the MIMO codes
to eliminate ISI caused by fading effects and its performance
can be analyzed. As such, channel estimation involving
training symbols and preambles can be applied to estimate
the channel instead of the ML channel estimation used in
this paper.

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]
VII. REFERENCES
[15]

[1] K.Pandey and M.Saad, "Modelling of indoor power

line for home application using broadband,"
International Journal of Enhanced Research in
Science Technology & Engineering, vol. 3, no. 4, pp.
173-178, April 2014.
[2] L.T.Berger, A.Schwager and J.J.Escudero-Garzas,
"Power Line Communications for Smart Grid

[16]

10

Applications," Journal of Electrical and Computer

Engineering, 2013.
J.A.Corts, L.Dez, F.J.Caete and J.J.SnchezMartnez, "Analysis of the Indoor Broadband PowerLine Noise Scenario," IEEE Transactions on
Electromagnetic Compatibility, vol. 52, no. 4, Nov
2010.
B.Rajkumarsingh and N.S.Poonye, "Modelling of
Power Line Communication Channel for Automatic
Meter Reading System with LDPC codes," GSTF
Journal of Engineering Technology (JET), vol. 3, no.
1, July 2014.
M. Bhatnagar, A. Mathur and B. Panigrahi, "PLC
Performance Analysis Over Rayleigh Fading Channel
Under Nakagami-m Additive Noise," IEEE
Communications Letters, vol. 18, no. 12, Dec 2014.
H.B.Celebi, "Noise and multipath characteristics of
power line communication channels," Graduate
Theses and Dissertations, Florida, 2010.
B.Tan
and
J.S.Thompson,
"Powerline
Communications Channel Modelling Methodology
Based on Statistical Features," Mar 2012.
M.Zimmermann and K.Dostert, "A Multi-Path Signal
Propagation Model for the Power Line Channel in the
High Frequency Range".
P.Mlynek, J.Misurec, M.Koutny and P.Silhavy,
"Two-port Network Transfer Function for Power Line
Topology Modelling," Radio Engineering, vol. 21,
no. 1, April 2012.
V.P.Singh, "Analysis of Power Line Communication
Channel Model using Communication Techniques,"
North Dakota, May 2012.
M. B. A.Mathur, "PLC Performance Analysis Over
Rayleigh Fading Channel Under Nakagami-m
Additive Noise," IEEE Communications Letters, vol.
18, no. 12, Dec 2014.
D.Duche and V.Gogate, "Signal Attenuation in
PowerLine Communication Channel," International
Journal of Emerging Trends & Technology in
Computer Science, vol. 3, no. 2, April 2014.
M.Zimmermann and K.Dostert, "An Analysis of the
Broadband Noise Scenario in PowerLine Networks,"
2006.
E.Liu, Y.Gao, O.Bilal and TKorhonen, "Broadband
Characterization of Indoor Powerline Channel," vol.
2, no. IEEE Transactions, pp. 901-905, May 2005.
A.D.Familua, "Noise Modelling for Standard Cenelec
A-Band Power Line Communication Channel,"
Johannesburg, Feb 2012.
H.Meng, Y.Guan and S.Chen, "Modelling and
analysis of noise effects on broadband power-line
communications," IEEE Transactions on Power
Delivery, vol. 20, no. 2, pp. 630 - 637, 2005.

[17] M.Arzberger,
K.Dostert,
T.Waldeck
and
M.Zimmermann, "Fundamental Properties of the Low
Voltage
Power
Distribution
Grid,"
Proc.
International
Symposium
on
Power-Line
Communications and Its Applications (ISPLC), 1997.
[18] L.T.Berger, A. P.Pagani and D.M.Schneider, "MIMO
Power Line Communicatons," IEEE Communications
Surveys & Tutorials, 2013.
[19] S.D.Santumon and B.R.Sujatha, "Space-TIme Block
Coding
(STBC)
for
Wireless
Networks,"
International Journal of Distributed and Parallel
Systems (IJDPS), vol. 3, no. 4, July 2012.
[20] S.Alamouti, "A simple transmit diversity technique
for wireless communications," IEEE Journal on
Selected Areas In Communications, vol. 16, pp. 14511458, Oct 1998.

[21] R.Kumar and R.Saxena, "Performance Analysis of

MIMO-STBC Systems with Higher Coding Rate
Using Adaptive Semiblind Channel Estimation
Scheme," The Scientific World Journal, Feb 2014.
[22] A.Papaioannou,
G.D.Papadopoulos
and
F.N.Pavlidou, "Performance of Space-time Block
coding over the power line channel in comparison
with the wireless channel".

11